Re: [lwip-devel] RFC3542 ICMPv6 Checksum Computation for Raw Sockets

[Ivan Delamer]

Thank you Grant!

I'm more or less on holidays now so I will look into merging this in January.

Cheers
Ivan


On Thu, December 20, 2012 09:38, Grant Erickson wrote:
> On Dec 19, 2012, at 3:30 PM, Ivan Delamer wrote:
>> There hasn't been a discussion about this yet. Feel free to propose one,
>> or even better, a patch that implements it. I can review it and work it
>> into the code if suitable.
>
> Ivan,
>
> Patches 7904 <https://savannah.nongnu.org/patch/index.php?7904> and 7905
> <https://savannah.nongnu.org/patch/index.php?7905> have been submitted.
> With these in place, a rudimentary ping6 to a link-local address works:
>
>       > ping6 -c 4 fe80::ea06:88ff:fecc:a9f4
>       PING fe80::ea06:88ff:fecc:a9f4 (FE80::EA06:88FF:FECC:A9F4): 56(104) 
> bytes
> of data
>       64 bytes from fe80::ea06:88ff:fecc:a9f4 (FE80::EA06:88FF:FECC:A9F4):
> icmp_seq=1 ttl=64 time=0.000 ms
>       64 bytes from fe80::ea06:88ff:fecc:a9f4 (FE80::EA06:88FF:FECC:A9F4):
> icmp_seq=2 ttl=64 time=0.000 ms
>       64 bytes from fe80::ea06:88ff:fecc:a9f4 (FE80::EA06:88FF:FECC:A9F4):
> icmp_seq=3 ttl=64 time=0.000 ms
>       64 bytes from fe80::ea06:88ff:fecc:a9f4 (FE80::EA06:88FF:FECC:A9F4):
> icmp_seq=4 ttl=64 time=0.000 ms
>
>       --- fe80::ea06:88ff:fecc:a9f4 ping statistics ---
>       4 packets transmitted, 4 received, 0 duplicates, 0 errors 0% packet 
> loss,
> time 6042ms
>       round-trip min/avg/max = 0.000/0.000/0.000 ms
>
> -Grant
>
>> On Wed, December 19, 2012 10:21, Grant Erickson wrote:
>>> I am implementing a basic ping6 test application for LwIP using a:
>>>
>>>     socket = lwip_socket(AF_INET6, SOCK_RAW, IP6_NEXTH_ICMP6);
>>>
>>> raw LWIP socket and it's mostly working, with checksum computation
>>> excepted. Per RFC3542 <http://www.ietf.org/rfc/rfc3542.txt>, "3.1.
>>> Checksums", outside of link-local addresses, for which a priori
>>> knowledge
>>> is fairly straightforward, the source address for pseudo-header
>>> checksum
>>> computation isn't easily derived from the application socket.
>>>
>>> Have there been any strong opinions or directions toward supporting a
>>> raw
>>> IPv6 checksum computation ala a socket option?
>
>